TMT model working in v4.0

[KarlClauser]

Hi Robbin,

Great to hear that v4.0 is out. Has the TMT model format been updated to run in v4.0? Lack of v3.0 compatibility has left me stuck on v2.36. When I last checked with you ~1 year you indicated that TMT model formatting was low priority for you. We run 90% unlabeled and 10% TMT, but the TMT support is a must have for our group.

Thanks,

--Karl

[RobbinBouwmeester]

Hi Karl,

Apologies for not responding sooner. Currently the training set does not contain TMT peptides. However, I will include some TMT data soon. Hopefully these will be included over the coming weeks!

Kind regards,

Robbin

[vrkosk]

I agree that it would be really nice to have TMT support in the latest version. We have hesitated updating from DeepLC 2.x for the same reason, even though DeepLC 3.x (and probably 4.0) is superior in every other way.

[RobbinBouwmeester]

Hi Karl and Vile,

Well, there is good news ;). The latest version (>= v4.1.0) supports TMT!

Due to the multitask design you do not have to select a specific TMT model as it has trained on all available data at once. The API has changed a bit so please take a look at that.

Kind regards,

Robbin

[RobbinBouwmeester]

Please note that the default model has now changed to the new architecture that supports the TMT (v4.1.1). Previously it defaulted to the legacy model, so please use v4.1.1 if testing on TMT